AI-Found Bugs Aren't Proving Any Easier to Exploit Despite the Hype

VulnCheck reports that AI-assisted vulnerability discovery has not increased real-world attacks, with only 1.3 percent of 1,061 discoveries exploited. This rate mirrors general vulnerability trends, contradicting claims that frontier AI creates instantly weaponizable bugs. AI currently increases bug volume rather than ease of weaponization, specifically regarding Anthropic's Project Glasswing. Separately, Google analysis of 15 million AI interactions indicates that most tasks across various jobs remain unaffected, showing workers are not automating themselves away.